Today, I'm Carl Adams and I'm here because of Dave's incredible generosity in sharing this channel with folks like me. So if you're not familiar with my channel and its content, audio is my thing. Do a lot of audio projects. I'm a bit of a nostalgic tragic.

You'll see that my choice to test equipment and I've got the sort of leaning towards analog techniques that you'd expect from someone whose day job is as a software engineer. so enjoy! So today I thought I'd look at some techniques for measuring distortion in audio amplifier, particularly techniques for trying to measure very low levels of distortion. I've got a little project underway that'll amplify based on the LM 1875, but in a compound configuration with it inside the feedback loop of another Op amp which promises to give some impressively good distortion numbers, but then then is challenging to measure on an instrument like via 3562 I which has 75, maybe 80 DB of dynamic range tops. So I was going to experiment with using an old-fashioned analog distortion meter as a notch filter basically as the front end prior to the signal into the DSi.

Let's give it a go. So one of the issues with trying to do accurate to distortion measurements instrument like 3562 A is that we can't really use the internal source of the 3562 A because it's not quite clean enough for our purposes. So what I've gone with is this loot here which it looks like I made it myself, but truth is I didn't actually bought the board off eBay a few years ago from a chap in Latvia I believe he made some very fine oscillator boards I Forget exactly how good the performance in terms of distortion these things are supposed to have, but there's a lot of zeros involved I know that much so it's certainly going to be better than our ability to measure, so that should be fine. And just as a initial test, I've just led that straight into the 3562 eye to see whether we can see any trace of any harmonics at all.

So we're doing a harmonic distortion measurement of our source just straight into the 3562 eye and we're covering a frequency spans from 100 Hertz to 5.1 K and we set the fundamental to 996 Hertz Because that's what our source is actually putting out, as we need to make sure that the analyzer does actually correctly identify the harmonics. but even with some averaging on pretty much all the harmonics are right down there in the grass. There's basically nothing to be seen there. so we're seeing 86 87 DB Basically, that's just the limitation of the instrument.

We're not able to measure really any harmonics at all. so I try something a little different. So what we've done here is we've now led the output of our source into the distortion meter. So this is a leader Brain distortion meter from the mid-1980s I would believe I Previously actually had the model before this, the Ldm 170, which unfortunately didn't have the the automatic knowing mode that this one does and certainly trying to use that and it's originally intended purpose was most frustrating as attempting to get the not filter properly null and keep it that way was very tricky indeed.

Anyway, so our output is going to be approximately one volt. RMS and into a level setting mode, just adjust the level set to try and get that. Come on, do that without sticking your head in front of a camera. It would be good.

There we go. Okay, so we now have the little set and we can start adjusting. This is actually pretty good. So frequency is set pretty close where it should be.

it's 10 times 100 around about one kilohertz and we can just tweak the yeah it's probably quite close off and use this at a similar sort of frequency so we can start making our way down the line. Pretty good frequency adjustment. It's not really changing anything at that point. We should probably leave it alone.

Okay, just play with the coarse balance at that point. Like just touch that and see if they can get better all catwalk. Okay, that's good making your way down, huh? dear? Right to touch the coarse adjustment. Yeah.

I Still good. Oh, look at that now. Yes. I Don't worry I want to try touching that now I Don't think we'll try that one.

Are you doing anything alright? let's go to there and it's a manual mode. I Wanted To start out in order. let's see if I can do any fit over any like Oh oh and I made it worse. Here we go and yeah, Oh probably shouldn't touch that, but go near it, see what happens that's good.

Oh, here we go. Yeah, that's good. Alright now I'll whack it in auto. Ok, so we're currently in a mode where 1% distortion is a full scale.

So if you're not familiar with these instruments, basically they're a notch filter followed by amplifier and an AC milli voltmeter. So basically anything that's left over after the not sugar is considered to be the THD plus in. Now the thing about this is that the limiting factor for these things is always the depth of the notch. So I mean this thing's best measurement range is 0.1% full scale and I think it's point zero One five or something like that is about the minimum reading you'll ever get off of it.

But that's not because it's point zero one five percent distortion inside the Machine Simply that it cannot suppress the fundamental any more than now with our dynamic signal analyzer and a limiting factor there is that it has a 14 bit analog to digital converter. So although the the DSA has enormous dynamic range in terms of be able to deal with input signals of different magnitudes, in terms of dealing with signals that are occurring at the same time, it's limited by the number of bits in the ADC, so that's the reason why we can't measure vanishingly small amounts of distortion. However, because we have output terminals on here, we can actually look at the signal that's coming after the notch filter. So by using this notch filter to remove the by far the largest part of it, you may input signal.

We can make the the job of the DSA much easier. Don't only that, because the the DSA allows us to not just measure the the THD in terms of measuring Lis harmonics in proportion to the fundamental. It also has a mode that can measure harmonic power, and by using that mode instead, we can actually directly read THD in conjunction with the settings here. So okay, so we're currently in the Minus 40 DB range, so we would expect that full scale which is one volt RMS would correspond to minus 40 dB.

Now if we now just flick the lights out and have a look at what's happening on the display signal analyzer, you'll see that we are getting a harmonic power of minus Sixty, Six Point, Six point, seven, dB. So what we need to do is take another 40 DB from that. So that's around about minus 106 point 7. In this case, we're measuring in THD over the first first five harmonics, just were just looking at a small part of the spectrum there.

We could look a bit further but already matter. I Just wanted that to be nice and clear on the screen. As you can see the noise floor starts to rise so we do have some issues there. You might think with a a device that can measure down to minus 80 dB not filtering into a device that can also measure down - a DB that you get some whiz-bang - 160 DB of dynamic range, but it doesn't work that way.

The noise floor and probably the residual distortion the analog distortion meter start to really head. No. Nevertheless, we're we're looking much more deeply into the distortion spectrum of our source and that's looking quite promising. So I think that illustrates how we could perform the same thing on a device under test.

I Hope you enjoy this little demonstration How with some inexpensive equipment - and a bit of imagination, you can improve your distortion measurements by about a factor of 10. Thanks for watching.